Evoked transients of pH-sensitive fluorescent false neurotransmitter reveal dopamine hot spots in the globus pallidus

  1. Jozsef Meszaros
  2. Timothy Cheung
  3. Maya M Erler
  4. Un Jung Kang
  5. Dalibor Sames
  6. Christoph Kellendonk  Is a corresponding author
  7. David Sulzer  Is a corresponding author
  1. Mortimer B. Zuckerman Mind Brain Behavior Institute, Columbia University, United States
  2. Columbia University, United States
  3. College of Physicians and Surgeons, Columbia University, United States
  4. Division of Molecular Therapeutics, New York State Psychiatric Institute, United States

Abstract

Dopamine neurotransmission is suspected to play important physiological roles in multiple sparsely innervated brain nuclei, but there has not been a means to measure synaptic dopamine release in such regions. The globus pallidus externa (GPe) is a major locus in the basal ganglia that displays a sparse innervation of en passant dopamine axonal fibers. Due to the low levels of innervation that preclude electrochemical analysis, it is unknown if these axons engage in neurotransmission. To address this, we introduce an optical approach using a pH-sensitive fluorescent false neurotransmitter, FFN102, that exhibits increased fluorescence upon exocytosis from the acidic synaptic vesicle to the neutral extracellular milieu. In marked contrast to the striatum, FFN102 transients in the mouse GPe were spatially heterogeneous and smaller than in striatum with the exception of sparse hot spots. GPe transients were also significantly enhanced by high frequency stimulation. Our results support hot spots of dopamine release from substantia nigra axons.

https://doi.org/10.7554/eLife.42383.001

Introduction

Dopamine neurotransmission plays important roles in motor behavior, memory consolidation, reward prediction error and many other functions (Matthews et al., 2016; Takeuchi et al., 2016; Sharpe et al., 2017; da Silva et al., 2018). While the dopamine release from the dorsal and ventral striatal projections of ventral midbrain neurons is by far the most studied, there is also dopamine release from projections of neurons with cell bodies in the hypothalamus, dorsal raphe (Matthews et al., 2016) and the locus coeruleus (Kempadoo et al., 2016; Sharpe et al., 2017), and responses to dopamine in many regions, including the hippocampus (Rosen et al., 2015; Kempadoo et al., 2016), paraventricular thalamus (Clark et al., 2017; Beas et al., 2018), bed nucleus of the stria terminalis and the amygdala (Matthews et al., 2016). A range of approaches provide evidence for spatially-restricted, temporally-precise dopamine signaling (Yagishita et al., 2014; Howe and Dombeck, 2016; Bamford et al., 2018). New optical approaches using fluorescent false neurotransmitters (FFNs) indicate that many dopaminergic axonal varicosities are functionally silent (Pereira et al., 2016; Liu et al., 2018), which may depend on the localization of molecular scaffolds (Liu et al., 2018), or presynaptic modulation by other neurons (Threlfell et al., 2012).

The GPe plays a significant role in shaping behavioral programs and initiating movement (Mallet et al., 2016; Mastro et al., 2017). Within the external globus pallidus (GPe), sparse dopamine axons were visualized in classic work using the glyoxylic acid method, which renders dopamine fluorescent in fixed tissue (Lindvall and Björklund, 1979). Branches of individual substantia nigra compacta (SNc) axons were visualized using a single cell viral infection protocol, demonstrating a relative paucity of axonal collaterals in the GPe (Matsuda et al., 2009).

It has not been known if sparse dopaminergic axons in GPe engage in dopamine neurotransmission. The level of dopamine release in the GPe is too sparse for analysis by carbon fiber amperometry and cyclic voltammetry, the principal methods for analyzing dopamine release in the heavily innervated dorsal and ventral striatum (nucleus accumbens). While very low levels of dopamine near detection threshold have been reported in the GPe in vivo by accruing a microdialysis sample over tens of minutes (Hauber and Fuchs, 2000; Fuchs and Hauber, 2004; Hegeman et al., 2016), there is no means to ascertain if this is due to GPe intrinsic neurotransmission or overflow from the highly innervated neighboring striatum.

We have an ongoing effort to develop FFNs to visualize neurotransmitter uptake and release in living brain tissue and in vivo. In this study, we adapt a pH-sensitive FFN, FFN102, which is a substrate for the dopamine transporter, DAT, and the vesicular monoamine transporter, VMAT2 (Rodriguez et al., 2013) as the first use of a ‘flashing FFN’. The signal from FFN102 is well-suited for studying synaptic release in sparsely innervated regions as it is brighter in the neutral extracellular milieu than the acidic milieu of the synaptic vesicle lumen (Lee et al., 2010). We find that upon electrical stimulation, FFN102 produces a calcium-dependent flash of fluorescence that requires functional dopamine fibers emanating from the SNc. The release properties differ from the striatum independently of DAT activity. These results indicate that pH-sensitive FFNs provide a means to study dopamine release within brain areas of sparse dopaminergic innervation.

Results

FFN102 release differs between the GPe and striatum

Fluorescent false neurotransmitters have previously been used to directly measure dopamine synaptic vesicle fusion and exocytosis from individual en passant release sites on the axon, termed puncta (Gubernator et al., 2009; Pereira et al., 2016). These FFN methods require tracking fluorescence within micron-sized regions in a field of view. Such methods use z-stacks to track puncta, which lowers temporal resolution. In these experiments, individual electrical pulses produce relatively small changes in fluorescence, and experimenters had to apply hundreds of pulses in order to generate a measurable signal.

As an alternate approach for sparsely innervated regions, we have adapted FFN102, a pH-sensitive fluorescent false neurotransmitter that is a substrate for the dopamine uptake transporter and the vesicular monoamine transporter and exhibits higher fluorescent emissia at extracellular neutral pH than the acidic synaptic vesicle pH (Lee et al., 2010; Rodriguez et al., 2013). As dopamine innervation in the GPe is very sparse, there are few FFN-labeled structures in a given field of view. We thus chose to average all pixels within each frame to provide a whole-field fluorescence measurement. In contrast to the endocytotic synaptic vesicle dye, FM1-43, FFN102 enters synaptic vesicles as a transporter substrate without electrical stimulation (Rodriguez et al., 2013). We thus used a 30 min incubation period without electrical stimulation to load cells with the probe (Figure 1A). To minimally disturb the synapse and allow for examining plasticity and modulation, we chose a stimulation paradigm that employed a brief stimulus period with an applied stimulus current of 200 μA. A bipolar electrode was placed on the slice and oriented so that the two poles contacted the dorsal and ventral aspects of the GPe. Fields of view imaged were 50 to 100 µm from one of the two electrode poles. We selected a stimulus frequency of 10 Hz, as it provided consistent responses and is within the range of dopamine neuron burst firing in vivo (Paladini and Roeper, 2014).

Electrical stimulation of GPe evokes FFN102 transients.

(A, B) Preparation of GPe brain slice. (C) In BAC-D2 GFP mice, the striatum and GPe are distinguishable, as the GPe receives a thick plexus of D2-positive terminals, while the striatum is rich in FFN labeled processes. Scale bar = 50 μM. (D) In response to 1 s long electrical stimulation at 10 Hz frequency, FFN fluorescence corresponding to regions in panel C) show a ‘flashing’ pattern of transients in the GPe (inset Roman numerals for traces ix-xv) and a prolonged and sustained increase in fluorescence in the striatum (inset Roman numerals for traces i-viii). Bars indicate the period of each stimulus (1 s). (E) Signal amplitude of stimuli at one interval in the GPe (from data in panel D), and a range of intervals in the striatum (from a different experiment): note the high variability of signal in GPe.

https://doi.org/10.7554/eLife.42383.002

To determine if the FFN release was localized within the GPe, and not due to diffusion from the striatum, we compared electrically stimulated changes in fluorescence within the striatum and GPe. To clearly delineate the boundary between the two areas, we used Drd2-BAC-GFP mice (Figure 1C): many D2 medium spiny neuron terminals converge within the GPe, creating a dense field of GFP fluorescence. While FFN102 labeled presynaptic elements profusely innervated the striatum, there were few obvious puncta within the GPe (Figure 1C, leftmost panel). When the slice was stimulated, the rapid alkalization of exocytosed FFN102 increased the whole-field fluorescence. The resulting fluorescence intensity profiles for these regions of interest are shown as red/pink lines and numbered i-viii for imaging within the striatum, and as blue lines numbered ix-xv in the GPe.

We observed different kinetics of FFN transients in the two regions. First, in the striatum, the fluorescence intensity remained elevated over time, while in the GPe the signal was much smaller and much shorter in duration. Second, in the striatum, pairs of stimuli needed to be separated by at least 60 s for the second transient to recover a signal comparable to the first, similar to the rate of recovery for dopamine release measured with cyclic voltammetry (Schmitz et al., 2002); in contrast, in the GPe, ten second intervals were often sufficient to recover the full response.

We performed control experiments to confirm if we were measuring release from dopaminergic terminals. First, as electrical stimulation can produce fluorescence changes through NADH metabolism in some in vitro preparations (Kasischke et al., 2004), we measured fluorescence changes in response to electrical stimulation in slices pre-incubated with or without FFN. To compare FFN transients, we averaged responses to multiple trains of stimuli in a field of view to obtain a slice-average FFN transient. Slices that had not been incubated with FFN showed no transient, measured as the area under the curve (AUC) (Figure 2A,B). Thus, FFN release, and not another process, was responsible for the transient events.

FFN102 and DAT dependence of fluorescence transients.

(A) Averaged traces of electrically evoked fluorescence from slices incubated with FFN, gold, or ACSF alone, blue. Traces reflect the average %ΔF/F, shading reflects ±SEM (n = 4 slices for ACSF, n = 6 slices for 10 μM FFN). (B) Box-and-whisker plot showing the AUC %ΔF/F evoked in the FFN and ACSF conditions. The edges of the boxes are 1 SEM from the mean, and the whiskers indicate ±2 SEM. Each point represents the average of images from one slice. ACSF slices were significantly different from FFN slices (two-tailed unpaired t-test, p < 0.01). For ACSF slices, the AUC %ΔF/F was 0.1 (CI95=[−0.78,0.98]) compared to 3.8 for FFN slices (CI95=[2.31,5.29]). (C) Transients from slices that were either incubated with FFN, gold, or FFN with 10 μM nomifensine, blue (n = 3 slices for FFN with nomifensine and n = 6 slices for FFN alone). (D) FFN102 transients evoked by electrical stimulation were strongly decreased by nomifensine (two-tailed unpaired t-test, p < 0.01; FFN, mean AUC %ΔF/F = 1.3 (CI95 = [0.61,1.99]) FFN with nomifensine, mean AUC %ΔF/F = 4.12 (CI95 = [2.97,5.27]).

https://doi.org/10.7554/eLife.42383.003

To examine if FFN102 is loaded in GPe axons as a substrate for the dopamine transporter (DAT), we co-incubated slices with both FFN102 and the DAT inhibitor, nomifensine, for 30 min before the stimuli. The significant decrease of evoked fluorescent transients in slices co-incubated with nomifensine is consistent with an inhibition of uptake and loading of FFN102 into DA axons (Figure 2C,D).

FFN102 transients reflect synaptic dopamine release

To assess whether the release of FFN102 within the GPe was of synaptic origin, we analyzed the effect of extracellular Ca2+ on FFN transient size. For these experiments, the levels of calcium reaching a slice were randomly alternated between 0.5, 2, and 4 mM Ca2+. An individual slice was perfused with a given concentration, allowing ten minutes for the calcium concentration to adjust within the slice. There was main effect of calcium levels on the AUC %ΔF/F (repeated measures ANOVA, p < 0.001) (Figure 3). Additionally, there was a significant difference between AUCs in the striatum and GPe (p < 0.001), as expected from our other experiments. We also found a significant interaction between calcium and region (p < 0.05). Consistent with release by synaptic vesicle fusion from dopamine axons in the striatum, 0.5 mM extracellular calcium produced significantly smaller average FFN102 striatal transients than 2 mM or 4 mM (Figure 3A,B). Similarly, within the GPe, calcium enhanced FFN102 transients (Figure 3C,D), with the response to increased calcium apparently saturating above 2 mM.

FFN102 transients are regulated by extracellular calcium.

(A) Average processed transients of electrically evoked transients from striatal areas in slices perfused with 0.5, 2, and 4 mM Ca2+ (n = 9 slices). (B) There was a main effect of calcium-level on AUC %ΔF/F (repeated measures ANOVA, p < 0.001). The AUC %ΔF/F was higher in 2 and 4 mM than 0.5 mM Ca2+ (two-tailed paired t-test, p < 0.001). (C) Average processed transients of electrically evoked transients from GPe in slices perfused with 0.5, 2, and 4 mM Ca2+ (n = 16 slices). (D) The AUC %ΔF/F was higher in 2 and 4 mM than 0.5 mM Ca2+ (two-tailed paired t-test, p < 0.001). Additionally, there was a significant interaction between calcium levels and region (p<0.05).

https://doi.org/10.7554/eLife.42383.004
Figure 3—source data 1

Fluorescence time courses for FFN transients evoked under varying calcium levels.

https://doi.org/10.7554/eLife.42383.005
Figure 3—source code 1

Analysis and figures for FFN102 transients under varying calcium levels.

https://doi.org/10.7554/eLife.42383.006

To confirm whether dopaminergic neurons were the source of the FFN transients, we used the toxin 6-hydroxydopamine (6-OHDA lesion) to unilaterally lesion the dopamine projections passing through the medial forebrain bundle. We used tyrosine hydroxylase (TH) immunolabel to confirm the lesions. We subsequently prepared slices from both hemispheres and recorded FFN transients within each slice (Figure 4A,B). The global averages of FFN transients for the lesioned and unlesioned sides were compared. Consistent with the hypothesis that FFN is released from DA neurons, 6-OHDA lesion significantly decreased the FFN transients (Figure 4B,C). After imaging FFN transients, we post-fixed and confirmed the lesion in the slices used for imaging. We noted that FFN transients in the non-lesioned side were decreased relative to control slices. Therefore, 6-OHDA depletion in one hemisphere may lead to a reduction of the GPe FFN transient on the contralateral side, consistent with reports that contralateral dopamine projections extend to the GPe (Pritzel et al., 1983; Douglas et al., 1987).

FFN102 transients are ablated in dopamine depleted mice.

(A) Representative images of slices in 6-OHDA experiments, post-fixed after imaging and immunolabeled for tyrosine hydroxylase. (B) Evoked transients from control side, blue, or lesioned side, green (n = 5 slices from five mice for each condition; shading represents SEM). (C) The lesioned hemispheres show decreased transients (average difference of 2.46, CI95 = [2.10, 2.82], two-tailed paired t-test p < 0.001). (D) Representative immunolabel for tyrosine hydroxylase in wild-type and aphakia mice shows a decreased label in the SNc, but not VTA. (E) Evoked transients wild-type mice of the same background, pink, aphakia mice, blue, and unstimulated aphakia mice, green (shading represents SEM). (F) Comparison of stimulation-evoked fluorescence changes in aphakia and wild-type mice (average difference in AUC %ΔF/F = 4.54 (CI95 = [3.05, 6.03], n = 5 slices from five mice for WT experiments, p < 0.001); between stimulated and unstimulated slices from aphakia mice, the average difference in the AUC %ΔF/F = 1.46 (CI95 of difference = [1.00, 1.91]; n = 3 slices from three mice for aphakia experiments; p < 0.05).

https://doi.org/10.7554/eLife.42383.007

To assess the anatomical source of the evoked transients, we turned to an aphakia mouse line, in which mutation of the pitx3 gene prevents the development of dopamine neurons selectively in the substantia nigra pars compacta (SNc) while sparing VTA dopamine neurons (Nunes et al., 2003; van den Munckhof et al., 2003). We confirmed this depletion in our mice using TH immunolabel (Figure 4D). To obtain within-animal comparisons for aphakia mice, we randomly interleaved unstimulated ‘sham’ imaging epochs during which we collected images without electrical stimulation. We found that wild type mice had significantly larger FFN transients than aphakia mice (Figure 4E). The aphakia slices showed a small but significant difference between stimulated and unstimulated experiments, possibly arising from spared VTA neurons (Figure 4F). Thus, SNc dopamine neurons are the primary contributor to FFN102 release in the GPe.

FFN102 release reveals differences in dopamine release between striatum and GPe

To examine the frequency dependence of FFN102 release in striatal and GPe dopamine synapses, we evoked FFN transients with 10 Hz and 50 Hz trains. We observed a main effect of frequency on FFN transients which was mostly due to the effects on GPe (two-factor ANOVA, p<0.001). Within the striatum, we did not observe significant differences in FFN transient size at 10 Hz and 50 Hz stimuli (Figure 5A). In contrast, within the GPe, the area under the curve for the transient evoked by 50 Hz was significantly higher than for 10 Hz (Figure 5B,C).

Modulation of FFN102 transients by stimulus frequency.

(A) FFN transients evoked in the striatum by five pulses (red lines) at 10 Hz, gold traces, or 50 Hz, dark gray traces (N = 9 slices for 10 Hz stimuli, N = 12 slices for 50 Hz stimuli; shading represents SEM). (B) FFN transients evoked in the GPe by five pulses at either 10 Hz, bright blue traces, or 50 Hz, dark blue traces (N = 52 slices for 10 Hz stimulation, N = 68 slices for 50 Hz stimulation). (C) The AUC %ΔF/F for the period from 0 to 300 ms from stimulus onset. Striatal responses were not significantly different at higher frequencies (CI95 of %ΔF/F at 10 Hz = [12.6,19.1] versus at 50 Hz = [14.7,26.5]), whereas GPe showed significantly higher AUC %ΔF/F at 50 Hz than 10 Hz (CI95 of %ΔF/F at 10 Hz = [3.46, 5.99] versus at 50 Hz = [9.28, 11.5]), significance assessed using two-tailed unpaired t-test, p < 0.001).

https://doi.org/10.7554/eLife.42383.008
Figure 5—source data 1

Fluorescence time courses for FFN transients evoked by 10 Hz and 50 Hz stimulation.

https://doi.org/10.7554/eLife.42383.009
Figure 5—source code 1

Analysis and figures for FFN102 transients evoked by 10 Hz and 50 Hz stimulation.

https://doi.org/10.7554/eLife.42383.010

We then addressed whether the differences in kinetics and size of FFN transients between the striatum and the GPe might result from differential reuptake, which might broaden the signal. For these experiments, we performed within-slice paired measurements of FFN transients in ACSF alone and after 10 min of perfusion with 10 μM nomifensine-containing ACSF. Nomifensine slightly decreased the FFN transient size in the striatum (Figure 6A,B), but not in the GPe (Figure 6C,D). Neither the decay time to its half-maximum value (Figure 6E) nor the shape of the decay were altered by nomifensine (Figure 6F), indicating a lack of a role for DAT in altering evoked FFN102 signals.

Differences in FFN102 transients in striatum and GPe are unrelated to reuptake.

(A) FFN transients imaged in the striatum for slices perfused first with ACSF alone followed by ten minutes with nomifensine. (B) The average AUC during the stimulus was 8.56 for control slices with a CI95 = [6.92,10.20], and for nomifensine, the average AUC was 5.90 with a CI95 = [3.94, 7.85] (paired t-test, p < 0.05). (C) FFN transients imaged in the GPe for slices similarly treated. (D) The average AUC during the stimulation time period was 4.16 with a CI95 = [2.56, 5.75], and for nomifensine, the average AUC was 2.78 with a CI95 = [0.81, 4.75] (paired t-test, p > 0.05). (E) Decay constants of log-transformed GPe traces show the time for the transient to decay to 10% of its initial value. Treated slices had an average decay time of 2.09 s, CI95= [1.56, 2.62], and untreated slices had an average of 2.20 s, CI95= [1.61, 2.79]. (F) Goodness-of-fit measurements for log-transformed traces.

https://doi.org/10.7554/eLife.42383.011
Figure 6—source data 1

Fluorescence time courses for FFN transients evoked under nomifensine block.

https://doi.org/10.7554/eLife.42383.012
Figure 6—source code 1

Analysis and figures for FFN102 transients during nomifensine block.

https://doi.org/10.7554/eLife.42383.013

The greater release of FFN102 in GPe at higher frequencies (Figure 5) suggested that GPe dopamine axons might have the capacity to generate large FFN transients. Indeed, we occasionally observed transients in the GPe of comparable size to striatum (Figure 7A). Surprisingly, GPe regions with large FFN102 transients did not exhibit obvious FFN102 puncta within the field of view (Figure 7B). We then examined if the large GPe transients were due to the presence of any puncta. To do so, we measured AUC values for transients and their correlation with the Canny Edge sum, a measure of high contrast edges of an image, specifically those around fluorescent puncta and neuropil (Figure 7C). We found very little correlation between GPe transient magnitude and Canny Edge values (R = 0.051) and found a similarly low correlation between GPe transient magnitude and initial image fluorescence (R = −0.062). Thus, neither the number of puncta nor the initial image fluorescence appeared to be related to the variability in FFN transient size.

Spatial and temporal characterization of FFN102 transients.

(A) A representative FFN transient from striatum, blue, and a ‘hotspot’ within the GPe, red. (B) Canny filtered masks were calculated from the average of each field of view’s baseline images, returning a pixel value of ‘1’ if the area has a high contrast and otherwise ‘0’. The fields of view that produced these transients are shown, both in their raw form and as a Canny edge filtered image. White scale bar = 10 µm. (C) For each field of view, the Canny edge sum (x axis) is displayed with the amplitude of its FFN transient (y axis). Correlation values are shown for GPe (n = 494 fields of view) and striatum (n = 84 fields of view). (D) An area of a representative brain slice containing the GPe. The two axes display distance in µm along the long axis of the GPe. (E) Points oriented to show imaging locations along the long axis of the GPe, with colors corresponding to their AUCs. Slices are shown from left to right, from more lateral to more medial slices. (F) Histograms showing the distance between pairs of all fields of view imaged within a slice. The distributions are split into four quartiles, where pairs in the top quartile both had the largest FFN transients compared to other pairs in the same slice. Similarly, pairs in the bottom quartile both had smaller FFN transients compared to the other 75% of pairs. (G) A plot of the derivative of the fluorescence intensity over time, averaged over the fields of view for each region (N = 84 for striatum, N = 494 for GPe). (H) Values of the derivative of the fluorescence intensity over time, for three intervals: 100 ms prior to stimulation, 100 ms after the first pulse, and 100 ms after the second pulse. Error bars represent the mean with CI95. All non-overlapping error bars were significant at p < 0.001. The mean value for the derivative at the first pulse was 0.86% for the striatum and 0.20% for the GPe.

https://doi.org/10.7554/eLife.42383.014
Figure 7—source data 1

Fluorescence time courses for spatial and temporal analysis.

https://doi.org/10.7554/eLife.42383.015
Figure 7—source code 1

Analysis and figures for FFN102 transients used for spatial and temporal analysis.

https://doi.org/10.7554/eLife.42383.016

We next examined whether the regions of high release were spatially clustered. We found that GPe slices rarely contained more than one such ‘hotspot’ of FFN release (Figure 7D,E). To quantify these observations, we segregated fields of view from a slice, assuming each field contains a putative release area, into four equal clusters based on the AUC from that field of view. We measured the distance between putative release areas (Figure 7F). If the hotspots were clustered, we would expect that the distribution of the top quarter of fields of view to have a small distance between pairs of release sites. We found that the most active areas were not closely spaced, and were as likely to be neighbors with a low releasing area as a high releasing area.

To compare the response to each electrical pulse between the GPe and striatum, we analyzed the derivative of the FFN transients, binned to average the intensity for frames corresponding to each electrical pulse. The averaged derivative for all regions of striatum and GPe was a sharp, single peak (Figure 7G), with a larger striatal peak due to the larger transient. We then determined each field of view’s derivative value before stimuli, after a single pulse, and after two pulses (Figure 7H). While we only resolve signals at 100 msec intervals, it is apparent that for both the striatum and the GPe, the largest contribution to the total transient occurred at the first electrical pulse and is consistent with estimates of the rising phase using cyclic voltammetry, which is ~180 msec in the striatum (Schmitz et al., 2001).

Discussion

A variety of FFNs now provide means to observe release during synaptic vesicle fusion from specific sites on axons, including striatal dopamine axons in vitro (Gubernator et al., 2009; Rodriguez et al., 2013; Pereira et al., 2016) and norepinephrine axons in vitro and in vivo (Dunn et al., 2018). Here, we introduce a new use for pH sensitive FFNs, which is to resolve release from dopamine axons in a region of very sparse innervation by ‘FFN flashes’, that is short duration calcium-dependent release transients during exocytosis as the fluorescence is unquenched upon exocytosis from the acidic vesicle lumen to the neutral extracellular space. FFN102, a DAT and VMAT2 substrate with a pKa of 6.2 (Lee et al., 2010), is well suited for this application, as its signal increases nearly 400% between the acidic synaptic vesicle (~pH 5.6) and the extracellular milieu at pH 7.4 (Rodriguez et al., 2013).

As might be expected from the sparse innervation by dopamine axons in the GPe, the same electrical stimulus generally elicited far smaller FFN transients in the GPe than the striatum, but occasional GPe ‘hotspots’, about one field of view per GPe slice, were found in which the level of evoked release was similar. The GPe hotspots had no clear lateral-to-medial pattern and were spatially segregated. Such hotspots of release have been described within the striatum as regions that include dopamine diffusing from active, but distant, sites of release (May and Wightman, 1989; Rodriguez et al., 2006). Alternatively, dopamine hotspots may be present where other modulatory cells are present, for example cholinergic cells, of which a few have been observed in the GPe (Gielow and Zaborszky, 2017). The identification of hotspots in the GPe is consistent with recent work showing that many striatal dopamine varicosities with clusters of synaptic vesicles are silent (Pereira et al., 2016), and this could be due to the absence or presence of local presynaptic scaffolding proteins (Liu et al., 2018). The GPe hotspots might represent areas where multiple collateral dopamine axons converge or areas with enriched presynaptic proteins. However, these active loci apparently do not contain large clusters of VMAT2-expressing vesicles observed after FFN102 loading. We also note that electrical stimulus of the striatum also stimulates striatal cholinergic interneurons, which induce additional release (Melchior et al., 2015), and that is expected to play far less of a role in the GPe, which is not known to have intrinsic cholinergic neurons. Local interactions with cholinergic interneurons may contribute to differences in the FFN signals between striatum and GPe, although these effects are minimized with the train stimuli used in this study (Rice and Cragg, 2004; Zhang and Sulzer, 2004; Cachope et al., 2012; Melchior et al., 2015).

While striatal FFN transients were easily measured in the striatum in response to a single stimulus, the much smaller FFN transients in the GPe were most effectively detected with a train of stimuli. In marked contrast to the striatum, the FFN transient in the GPe decayed rapidly (Figure 6E,F) consistent with diffusion from a small amount of release sites, and appeared as a ‘flash’. We used trains of 10 pulses to release sufficient FFN, which prolongs the signal. Most GPe regions showed release after the first pulse, but the amount of evoked release was strikingly variable, and far more responsive to increased stimulus frequency than release in the striatum. The amplitude of the GPe FFN transients was not correlated with the number of puncta in the images (Figure 7C). A recent ultrastructural study of dopamine axons in the striatum indicates that dopamine synaptic vesicles are present throughout the axon, with synaptic vesicle clusters at identifiable synapses (Gaugler et al., 2012). The lower amplitude evoked FFN transients in GPe may be due to exocytosis from comparatively small reservoirs of synaptic vesicles in thin portions of the axon that do not maintain large vesicle clusters. Indeed, the sparse observable puncta in GPe (i.e., apparent boutons with FFN loaded vesicles) are functionally silent (in the striatum,~80% of puncta are inactive under electrical stimulation). In the current work we discovered that the FFN release occurs primarily at sites that do not contain clusters of vesicles observable by FFN imaging.

The decay of FFN102 transients to the apparent detection limit was much faster in the GPe than the striatum, including in GPe hotspots with high evoked transient amplitudes similar to the striatum. The GPe and other sparsely innervated regions have little or no dopamine uptake transporter (DAT) activity (Miller et al., 1997). Differences in reuptake, however, were not responsible for the difference in transient kinetics, as while DAT blockade during FFN102 incubation blocked transients by inhibiting loading (Figure 2), DAT blockade during the stimuli did not affect the transients in either striatum or GPe (Figure 6). The longer fluorescence decay in striatum is in part due to diffusion from out-of-plane striatal terminals, whereas in the GPe, areas of high release are few and spatially separated, and so less signal would diffuse from distal release sites to the regions of interest (Sulzer and Pothos, 2000).

While tracing studies show that SNc neurons pass through the GPe (Matsuda et al., 2009), there could be additional sources of GPe dopamine release. For example, recent work highlights dopaminergic dorsal raphe cell inputs to the amygdala (Matthews et al., 2016) and locus coeruleus inputs to the hippocampus (Kempadoo et al., 2016). By specifically depleting SNc DA axons using toxins, and examining the aphakia mouse line in which SNc DA neurons are absent and dopamine release in dorsal striatum is abolished (Lieberman et al., 2018), we confirmed that the axons that release FFN102 are dopaminergic fibers that emanate from the SNc (Figure 4).

The use of pH-sensitive ‘flashing’ FFNs as optical analogs to cyclic voltammetry and amperometry for the detection of catecholamine neurotransmission in regions of low innervation may prove to be a useful experimental tool. The current approach is able to resolve FFN transients in 30 × 30 µm fields of view, close to the size of neuronal cell bodies, and so FFN transients might be used to characterize dopamine release near specific cell types, such as the arkypallidal cells which project back into the striatum and the protopallidal cells that project to the SNr and GPi (Gittis et al., 2014; Mastro et al., 2014; Hernández et al., 2015). Dopamine is known to disinhibit GPe cells (Cooper and Stanford, 2001; Shin et al., 2003), and could modulate the balance between the arky- and protopallidal pathways. The recently introduced FFN270, which is a substrate for the norepinephrine transporter (Dunn et al., 2018), is also pH sensitive and could provide means to measure properties of the sparse and widely distributed network of norepinephrine axons in the nervous system.

Materials and methods

Key resources table
Reagent type
(species)
or resource
DesignationSource or
reference
IdentifiersAdditional
information
Strain, strain
background
(Mus musculus)
Male and female
C57BL/6J mice
The Jackson
Laboratory
RRID:IMSR_JAX:000664
Strain, strain
background
(Mus musculus)
Male Drd2-
BAC-GFP mice
MMRCMGI:3843608
Strain, strain
background
(Mus musculus)
Male and female
aphakia mice
Provided by
Dr. Un Kang
(doi:10.1111/gbb.12210;
doi: 10.1073/pnas.1006511108)
Antibodymouse monoclonal
anti-tyrosine
hydroxylase
MilliporeRRID:AB_390204(1:750)

Ethics statement

Request a detailed protocol

All animal protocols followed NIH guidelines and were approved by Columbia University’s Institutional Animal Care and Use Committee (protocol AC-AAAR4420).

Mice

Experiments were performed on male and female C57BL/6J mice (9–24 weeks old) obtained from Jackson Laboratories (Bar Harbor, ME, USA), Drd2-BAC-GFP (S118Gsat/Mmnc) mice purchased from MMRRC, and aphakia mice. The aphakia allele is a loss-of function mutation in the Pitx3 gene arose spontaneously on the 129/Sv-S1 j strain at the Jackson Laboratory and was maintained in a C57BL background as previously described (Ding et al., 2007; Ding et al., 2011). All mice were housed under a 12 hr light/dark cycle in a temperature-controlled environment with food and water available ad libitum.

6-OHDA lesion

Request a detailed protocol

Unilateral 6-hydroxydopamine (6-OHDA) lesions were performed in male and female C57/BL6 mice of 12–15 weeks of age. Mice were anesthetized with ketamine/xylazine cocktail (80 mg/kg ketamine and 4 mg/kg xylazine, i.p.). Mice received unilateral lesion of the left medial forebrain bundle by intracranial infusion of 4.5 μg of 6-OHDA free base in 1.5 μl of 0.05% ascorbic in 0.9% saline at a rate of 300 s/μl into the following coordinates: anterior/posterior (−1.3 mm), medial/lateral (+1.2 mm), and ventral to skull surface (−5.4 mm) via a 28-gauge stainless-steel cannula that stayed in the brain for 5 min after the injection before being withdrawn. Desipramine (25 mg/kg, Sigma-Aldrich), a norepinephrine reuptake inhibitor, was injected intraperitoneally 30 min prior the infusion of 6-OHDA to protect norepinephrine neurons. Following surgery, mice received 2 weeks of intensive postoperative care consisting of twice daily, 1 ml injections (i.p.) of 5% dextrose in 0.9% saline and highly palatable, high fat content food (Bacon softies, Bio-Serv) as supplementation to the normal mouse chow diet. Mice were analyzed for FFN102 imaging two weeks after the unilateral lesion.

Imaging FFN102 transients

Request a detailed protocol

Sagittal slices containing the GPe were prepared as previously described with minor modifications (Pereira et al., 2016). Briefly, mice were killed by cervical dislocation and decapitated. Both male and female wild-type mice were used. A Leica VT1200 vibratome (Leica Microsystems, Wetzlar, Germany) was used to cut three 250 µm thick slices from each hemisphere. Slices were maintained at room temperature in oxygenated (95% O2, 5% CO2) artificial cerebrospinal fluid (ACSF [in mM]: 125.2 NaCl, 2.5 KCl, 26 NaHCO3, 0.3 KH2PO4, 2.4 CaCl2, 1.3 MgSO4, 10 glucose, 0.8 HEPES, pH 7.3–7.4, 295–305 mOsm), and used within 1 to 5 hr. Prior to imaging, each slice was incubated in an ACSF solution containing 10 µM FFN102 for 30 min. Slices were then transferred to a QE-1 imaging chamber (Warner Instruments) and held in place with a custom-made platinum wire and nylon holder. Slices were perfused with ACSF at a rate of 2–3 ml/min at room temperature (23°C).

All images were acquired using a Prairie Ultima Multiphoton Microscopy Systems (Bruker/Prairie Technologies) equipped either with a Spectra-Physics MaiTai HP DeepSee titanium-sapphire pulsed laser (Newport) and either a 10 × 1.0 NA air objective or 60 × 0.9 NA water immersion objective (Carl Zeiss Microscopy). For electrical stimulation, a twisted bipolar platinum stimulating electrode (Plastics One) was placed directly on top of the region to be imaged and pulses generated by an Iso-Flex stimulus isolator triggered through a Master-8 (each pulse 600 μs × 200 μA). Imaging was synchronized to the pulse generation using the TriggerSync software provided with the Prairie imaging system. FFN102 was excited at 760 nm and detected at 440–490 nm. For all imaging sessions, the stimulating electrode was placed using the 10x objective and then a region, approximately 50–100 μm from electrode tip, was examined under 60x at 10x digital zoom. Regions were only stimulated if they were 30 μm beneath the surface of the slice. Images, 64 × 64 pixels at 10x digital zoom (approximately 50 × 50 µm), were recorded at 10 Hz (6 µs dwell time) using a spiral scan. The use of a stimulus train provides a longer duration optical signal, and so a more robust detection of FFN release events by distinguishing the signal from rapid fluctuations. To minimize movement and deformation of slices, experiments were performed at room temperature.

Electrical stimulation protocols for brain slices

Request a detailed protocol

Electrical stimulation was synchronized with frame acquisition using TriggerSync software including with the PrairieView proprietary imaging package. Electrical stimulation protocols were chosen based on the type of experiment and are detailed and justified for each experiment.

Quantification of FFN102 fluorescence intensity

Request a detailed protocol

To analyze optical data, images were loaded into MATLAB and ΔF/F were extracted by first calculating the mean intensity of every image frame. To calculate baseline fluorescence, a linear fit was calculated to 500 ms of the data immediately prior to the stimulation. For each image sequence, this calculated baseline was subtracted from the fluorescence intensity over time and the resulting value divided by the baseline. This approach removes baseline activity due to bleaching or the washing out of FFN molecules. Fluctuations in baseline activity are much slower than the rise of an FFN transient, which are present at the first acquisition following the electrical stimulus. The smoothed baseline was used as the F for calculating ΔF/F for every time point of a given intensity. To calculate area under the curve values (AUC), we used the trapezoidal integration formula as implemented by MATLAB and measured the integral over frames within the stimulus period.

Measurement of FFN in regions of interest

Request a detailed protocol

Previous approaches to identify FFN labeled puncta relied on localizing the centers and boundaries of puncta (Pereira et al., 2016). In the case of flashing FFNs, which become brighter in the extracellular space, the unit of analysis is a field of view rather than individual puncta. We classify each pixel as belonging to the edge of an FFN-filled region or background by applying a Canny-edge filter using MATLAB’s canny function with a low threshold of 0.08 and a high threshold of 0.2. These values successfully outline the edge fluorescence from striatal images in which puncta are visible to the eye. The low threshold serves to extract even weak edges that may be present for thin varicosities or puncta with small quantities of FFN.

Measures of calcium and DAT dependence

Request a detailed protocol

For Ca2+ manipulation, slices were stimulated by using a single train of electrical pulses (10 pulses per train, 10 Hz) and imaged. The concentration of Ca2+ was varied by switching the perfusate between 0.5, 2, and 4 mM CaCl2, with a randomized order, with ten minutes between recording sessions to allow the calcium to perfuse into the slice. For the striatum, one field of view was imaged at the three concentrations and care was taken to ensure the field of view did not move during the ten minutes as the perfusate was switched. For the GPe, the same three fields of view were visited at each concentration and stimulated. The image intensity over time was obtained for each of the three stimulated fields of view, at each concentration, and an average calculated.

To measure DAT dependence in the striatum, one field of view was imaged in each slice and then nomifensine (10 μM, Sigma-Aldrich) dissolved in ACSF was perfused for ten minutes. The same field of view was then imaged with nomifensine-containing ACSF. For the GPe, ten fields of view were chosen at random, imaged, and then switched to nomifensine-containing ACSF. The same ten fields of view were visited and imaged with nomifensine-containing ACSF.

Histology

Request a detailed protocol

For 6-OHDA experiments, mouse brain slices were first used in FFN102 experiments. Slices were then removed from the ACSF perfusate and placed into ice-cold 4% PFA in 0.1M TBS and stored overnight at 4°C. Slices were washed for an hour, six times, in PBS. Slices were then blocked for 2 hr with 10% fetal bovine serum, 0.5% bovine serum albumin in 0.5% TBS-Triton X-100. Primary antibody against tyrosine hydroxylase (AB152, Millipore) was applied at 1:750 dilution in block for 72 hr at 4°C. Slices were washed again for an hour, six times. Secondary anti-rabbit antibody was applied in block for 16 hr at 4°C.

For the Pitx3 (aphakia) mouse SN and VTA histology, mice under deep anesthesia were transcardially perfused with ice-cold 4% PFA in 0.1M TBS. Brains were post-fixed overnight and washed in TBS for fifteen minutes, four times. Sections of 30 µm thickness were obtained using a Leica VT2000 vibratome. The sections were incubated in blocking solution for 1 hr at room temperature, then placed in blocking solution containing primary antibody to TH at 1:750, and incubated overnight at 4°C. The sections were washed again and incubated with anti-rabbit secondary antibody in blocking solution for 1 hr at room temperature. Images for the 6-OHDA and aphakia experiments were acquired at 2.5x using a Hamamatsu camera attached to a Carl Zeiss epifluorescence microscope. MATLAB was used to process the images for level adjustment.

Experimental design and statistical analyses

Request a detailed protocol

For experiments showing that FFN102 is necessary for evoking fluorescence transients (Figure 2), AUCs of fluorescence were measured for each slice, then slices from the two conditions were compared for significance using a t-test, and confidence intervals are also provided for comparison. To determine calcium dependence (Figure 3), within-slice comparisons of FFN transients were made at three separate concentrations. ANOVA was used to show dependence on calcium with each group comprising the average AUCs from a slice, obtained at different concentrations. Confidence intervals for the release at a given concentration were calculated based on the data for each group. To examine the effect of 6-OHDA treatment on the size of FFN AUC transients (Figure 4), we recorded from lesioned and unlesioned sides of each treated animal and compared pairs of hemispheres for each animal using a pairwise t-test. The pairwise differences were also used to compute confidence intervals to estimate the effect of depletion on the FFN signal. For aphakia animals, pairwise comparisons were made within a slice under two conditions: stimulated and not stimulated. A pairwise t-test was used to compare the two conditions and confidence intervals indicate the magnitude of the differences obtained between the two conditions for the aphakia mice. The same experiment was performed using wild-type mice, and the confidence intervals were similarly calculated. The difference between stimulated and unstimulated represented the independent samples to be compared for the aphakia and wild-type mice. Confidence intervals were used to compare these differences. To analyze the features of the striatal and GPe FFN transients (Figures 57), confidence intervals were calculated using data from multiple fields of view across multiple slices and animals. We calculated confidence intervals for comparison between striatum and GPe. For all parametric statistical tests, data were deemed normally distributed.

Data availability

All data generated or analysed during this study are included in the manuscript and supporting files.

References

Decision letter

  1. Inna Slutsky
    Reviewing Editor; Tel Aviv University, Israel
  2. Eve Marder
    Senior Editor; Brandeis University, United States

In the interests of transparency, eLife includes the editorial decision letter and accompanying author responses. A lightly edited version of the letter sent to the authors after peer review is shown, indicating the most substantive concerns; minor comments are not usually included.

[Editors’ note: a previous version of this study was rejected after peer review, but the authors submitted for reconsideration. The first decision letter after peer review is shown below.]

Thank you for submitting your work entitled "Using Fluorescent False Neurotransmitters to Characterize Exocytosis from Dopamine Synaptic Vesicles within the GPe" for consideration by eLife. Your article has been reviewed by two peer reviewers, and the evaluation has been overseen by a Reviewing Editor and a Senior Editor. The reviewers have opted to remain anonymous.

Our decision has been reached after consultation between the reviewers and the Reviewing Editor. Based on these discussions and the individual reviews below, we regret to inform you that your work will not be considered further for publication in eLife.

Although the reviewers appreciate the development of the tools to measure dopamine release in a sparsely innervated GPe area, the method, in its current form, is not sufficiently mature to characterize the properties of dopamine release and its modulation. Significant further work is needed to improve the signal-to-noise ratio of the indicator and to solidify the conclusions on the differential regulation of dopamine release in GPe versus striatum. See below to see the reviewers' comments in detail.

Reviewer #1:

In the manuscript by Meszaros et al., the authors describe their use of FFN102, a fluorescent analogue for dopamine, to characterize synaptic activity of dopaminergic neurons in the external globus pallidus. The challenge in studying dopaminergic presynaptic activity in this brain area is the scarcity of relevant synaptic terminals, which limits the applicability of other research techniques, such as cyclic voltammetry.

The authors' approach is to load brain slices with FFN102, which is loaded specifically into vesicles in dopaminergic presynaptic terminals. After washing away the extracellular FFN102, they electrically stimulated the slice and examined, using two-photon microscopy, the increase in fluorescence within two adjacent brain areas – the aforementioned GPe and the striatum, which is much more densely innervated by dopaminergic axons/terminals. The authors find that bursts of stimuli resulted in a transient increase in fluorescence in the GPe, and a much more long-lasting one in the striatum. To show that the increase is indeed the result of synaptic release of FFN102, the authors did the following controls: (1) They stimulated slices which had not been loaded with FFN102. (2) They stimulated FFN102-loaded slices in the presence of cadmium. (3) They stimulated slices incubated in various concentrations of extracellular calcium. The authors also: (4) lesioned dopaminergic neurons using 6-OHDA in one hemisphere, and (5) used a genetic model in which dopaminergic neurons fail to develop in the SNc. FFN102 fluorescence was examined in the lesioned area or in the SNc of aphakia mice.

Upon concluding that FFN102 can reveal synaptic dopaminergic activity, the authors examined differences in synaptic properties in the GPe and striatum, by (1) examining in more detail the fluorescence dynamics during the delivery of stimulation trains and (2) by examining the effect of the frequency of stimulation on the quantity of release. Finally, they examined modulation of release by D2, opiod and nicotinic receptors in the GPe. They concluded that release in the striatum is highly depressed (i.e. that it most of the difference in fluorescence is obtained during the first few stimuli), but that in the GPe it is less so (it depressed more slowly). Furthermore, GPe responses were enhanced by high-frequency stimulation. Finally, they showed that GPe responses responded to inhibition of cholinergic transmission, suggesting cholinergic modulation of presynaptic release.

1) My most significant concern relates to the assumption/claim of the authors concerning the nature of the fluorescence transient that they image. The authors indicate that fluorescence levels are integrated across the whole image, because the nature of release of FFN102 is such that fluorescence increases are not well-localized, probably because FFN102 diffuses after release. However, the authors also suggest: "FFN102 has slow reuptake kinetics and reacidification must also occur for the fluorescence to be quenched to background levels, a process that may require minutes in a heavily innervated region like the striatum." My issue with this assessment is that if the assumption is that FFN102 is released from the terminals and then diffuses into the tissue (as suggested by the need to integrate fluorescence) then I would not expect that the eventual decrease in fluorescence in the GPe would be strongly impacted by reuptake – as is implied in this paragraph. Because FFN102 is recognized specifically by DAT and VMAT, I would rather expect GPe fluorescence not to decline in this manner – because presynaptic terminals expressing DAT in this area are quite sparse (unless there is significant DAT in other locations). To address this concern, the authors could examine the effect of the acute inhibition of DAT on the GPe transients. I stress that I do not refer to experiments similar to those already reported by the authors in previous publications (Rodriguez et al., 2013), in which they showed that DAT inhibition stops the initial loading of FFN102. Rather, I would suggest inhibiting DAT in already-loaded slices and to examine the effect on the fluorescence transients after stimulation. If the difference in the fluorescence dynamics between GPe and striatum is indeed related to reuptake, I would expect DAT inhibitors to have a large effect, mostly in GPe. In the same context – the decay time of the fluorescence of FFN102 in the various figures shown in the manuscript appears to differ (for example compare Figure 3C to Figure 2A). Is the rate of fluorescence decrease relevant to the point I raised here?

2) In continuation to this point, I would expect the baseline fluorescence level in GPe to decrease successively with each given train of stimulation, if FFN102 that is loaded into vesicles is secreted and lost. In Figure 1E it is clear that some experiments behaved in this way, while others did not. Could the authors attempt to examine their data while considering this possible interpretation? What is the proportion of sessions in which fluorescence is decrease by stimulation? Is the decrease indeed related directly to stimulation?

3) I wonder about the authors' claim that Ai9 ChR2 mice have been used, while no data is presented. The ChR2 mice have the potential to increase the signal-to-noise significantly.

4) What is the sensitivity of the permutation method to noise in the imaging results? I would think that if baseline recordings are noisy, then the permutation method may increase the fraction of tests that produce a "significant" answer. Is there a difference in the noise level in the various groups of experiments?

5) The text refers to panels and data in Figure 6 (subsection “FFN102 release reveals differences in dopamine release between striatum and GPe”, first paragraph) that do not exist in the figure itself. This should be fixed before I can form an opinion on this data and its interpretation.

Reviewer #2:

In this paper, Meszaros and colleagues establish that a pH-sensitive false fluorescent neurotransmitter, FFN102, can be used to measure dopamine release in a brain area with sparse dopamine innervation. This has been a limitation in the field, and this manuscript is undoubtedly an advance. However, because the signals are small (roughly 0.3% DF/F for the signal amplitude, Figure 2A) temporal and spatial resolution remain limited. Release from single puncta cannot be measured, and action potential trains are needed to measure a signal above background. This results in a need for analyses of AUC (as opposed to amplitude) and permutation (for modulation). Although these points limit enthusiasm somewhat, the manuscript does provide a first (and likely an important) step towards better understanding of roles for dopamine in brain areas with sparse dopamine innervation.

1) FFN102 loading results in a lot of signal outside of dopamine neurons (Figure 2E). This matches well with Figure 7, where little dopamine release is detected in areas with diffuse "loading", and this diffuse lower intensity signal is also present in the striatum (Figure 2E here and Figure 2 in PMID23277566). The most likely explanation is that there is background accumulation of the dye in non-dopamine neurons. This has to be discussed and better acknowledged in the text. Furthermore, Figure 2E overlap panels are very saturated in the blue=FFN channel, different from the non-merged images that show FFN only. More careful image editing, perhaps combined with a different color choice, should be applied.

2) The experiments that address release characteristics and modulation are unclear at this point: a) The modulation experiments suggest that there is not the typical modulation of opioid and D2 receptors, and ACh receptor blockage perhaps mildly decreases the FFN102 signal during a stimulus train. However, the controls in the striatum suggest that FFN102 signals behave somewhat different from previous studies. The simple prediction from published studies would be that the first amplitude is smaller in MEC, but that dopamine release is rapidly depleted during 50 Hz/50 stimulus trains whether or not AChs are blocked. The method does not have time and spatial resolution to test this. Hence, it is possible that the differences observed have more to do with dye diffusion, clearance, etc., which could be different between the brain areas, than with release modulation.

b) Results describe experiments to suggest that "release differs between the GPe and the striatum" (section caption). Generally, better terminology to express the distinction between release and the FFN102 signal is needed. The signal is a function of release, diffusion, reuptake and reacidification. It is possible that differences in the signal arise because diffusion, reuptake etc. are different between the two brain areas, and release per se, from a bouton, is not different except that much less FFN102 is released because of the sparsity of dopamine terminals in GPe.

The meaning of these experiments is currently less clear than what the model shown in Figure 10 expresses. Leaving the data in the manuscript is fine to hint at potentially different modulation, but it may be better to remove the model and focus on the point that signals in an area with sparse dopamine innervation can be detected, rather than making strong mechanistic claims about the release.

3) The calcium dependence is shown for N = 1 slice in Figure 3C, D, this is simply below acceptable standards.

4) In Figure 7, it would be more meaningful to show a correlation between the morphological appearance (panel A) and the FFN transient (panel B). As shown, Figure 7 does not establish that areas with sparse but highly fluorescent axons release more. Furthermore, the "high" panels in A poorly reflect that point compared to the "low" panels. A better analysis is necessary.

[Editors’ note: what now follows is the decision letter after the authors submitted for further consideration.]

Thank you for resubmitting your work entitled "Evoked transients of pH-sensitive fluorescent false neurotransmitter reveal dopamine hot spots in the globus pallidus" for further consideration at eLife. Your revised article has been evaluated by Eve Marder (Senior Editor), a Reviewing Editor, and three reviewers.

While the manuscript has been substantially improved, some remaining important technical issues, summarized below, need to be addressed before the final decision on the manuscript is made.

Major comments:

1) All the illustrations used multiple stimuli to evoke release and the rise in fluorescence is slow and sustained. Single stimuli show rapid rise and fall of dopamine using both voltammetry and the new sensors. Thus the present results appear to be quite different. The authors should isolate responses to single stimuli (for instance by initially using sparse stimulation), followed by strong stimulation to identify the events and loci.

2) One conclusion was that the fluorescent increase in the globus pallidus was smaller and declined more rapidly. This is not surprising given that the experiments were done at room temperature. If the reuptake process plays any role in the decline in fluorescence would be inactive so the decline would be strictly dependent on diffusion. It could be that the lower concentration of FFN102 measured in the GP dropped below the detection limit thus appearing to be more rapid. Even in the hot spots of the GP the diffusion and dilution away from the point is most likely faster than might be expected from what is observed from multiple release sites in the striatum.

3) The difference in the results obtained with the paired stimulation in the GP and dorsal striatum could be accounted for by the presence of the cholinergic interneurons in the striatum that increase the probability of dopamine release dramatically such that there is substantial paired pulse depression of dopamine release.

4) There was no evidence for recovery (decrease) in fluorescence of FFN102 after cessation of stimulation in the striatum (see Figure 1). Therefore, the discussion of the authors concerning the kinetics of recovery doesn't appear to be consistent with the data (Discussion, fourth paragraph).

5) The idea of "hot spots" of release: The data presented in Figure 7 are not convincing. The authors explain that data were measured as an average of fields of view. How does this relate to the idea of hotspots? Especially when the authors refer to structures which should be of smaller dimensions than the fields. For example, in the Introduction, the authors write: "The identification of hotspots in the GPe also supports recent work showing that many striatal dopamine varicosities with clusters of synaptic vesicles are silent (Pereira et al., 2016), and this could be due to the absence or presence of local presynaptic scaffolding proteins (Liu et al., 2018)." However, the dimensions of the so-called "hot spots", as measured in the current study, are significantly larger than what one would expect varicosities to be. Therefore, such a suggestion as to the biological basis for the hotspots does not appear plausible. Later, the authors write: "For example FFN transients could be used to locate dopamine release near specific cell types, such as the arkypallidal cells which project back into the striatum and the protopallidal cells which project downstream to the SNr and GPi (Gittis et al., 2014; Mastro et al., 2014; Hernández et al., 2015)." Are the dimensions of the hotspots small enough to find specific cells within slices? If the authors measure transients from whole fields, then this does not appear to be the case.

https://doi.org/10.7554/eLife.42383.019

Author response

[Editors’ note: the author responses to the first round of peer review follow.]

We have now conducted a year’s worth of additional experiments and analysis that we believe fully answers all of the critiques, and we request that you reconsider if this can be reviewed again by the reviewers. Both Reviewers however expressed insightful critiques/concerns, and so we have carried out substantial new experiments and revised the study as they recommended.

[…] 1) My most significant concern relates to the assumption/claim of the authors concerning the nature of the fluorescence transient that they image. The authors indicate that fluorescence levels are integrated across the whole image, because the nature of release of FFN102 is such that fluorescence increases are not well-localized, probably because FFN102 diffuses after release. However, the authors also suggest: "FFN102 has slow reuptake kinetics and reacidification must also occur for the fluorescence to be quenched to background levels, a process that may require minutes in a heavily innervated region like the striatum." My issue with this assessment is that if the assumption is that FFN102 is released from the terminals and then diffuses into the tissue (as suggested by the need to integrate fluorescence) then I would not expect that the eventual decrease in fluorescence in the GPe would be strongly impacted by reuptake – as is implied in this paragraph. Because FFN102 is recognized specifically by DAT and VMAT, I would rather expect GPe fluorescence not to decline in this manner – because presynaptic terminals expressing DAT in this area are quite sparse (unless there is significant DAT in other locations). To address this concern, the authors could examine the effect of the acute inhibition of DAT on the GPe transients. I stress that I do not refer to experiments similar to those already reported by the authors in previous publications (Rodriguez et al., 2013), in which they showed that DAT inhibition stops the initial loading of FFN102. Rather, I would suggest inhibiting DAT in already-loaded slices and to examine the effect on the fluorescence transients after stimulation. If the difference in the fluorescence dynamics between GPe and striatum is indeed related to reuptake, I would expect DAT inhibitors to have a large effect, mostly in GPe. In the same context – the decay time of the fluorescence of FFN102 in the various figures shown in the manuscript appears to differ (for example compare Figure 3C to Figure 2A). Is the rate of fluorescence decrease relevant to the point I raised here?

We thank the reviewer for this insightful analysis, and agree that as a DAT substrate, it could be that reuptake might acutely inhibit the decay of the signal or increase its amplitude, and the decay times are indeed relevant to these points.

We therefore conducted additional experiments in which we acutely apply nomifensine into the FFN102 loaded slices We found that DAT inhibition neither increased the amplitude nor decay of the GPe transients. Thus, the FFN “flash” is a measure of release but not of acute reuptake. The new data are displayed in Figure 6.

2) In continuation to this point, I would expect the baseline fluorescence level in GPe to decrease successively with each given train of stimulation, if FFN102 that is loaded into vesicles is secreted and lost. In Figure 1E it is clear that some experiments behaved in this way, while others did not. Could the authors attempt to examine their data while considering this possible interpretation? What is the proportion of sessions in which fluorescence is decrease by stimulation? Is the decrease indeed related directly to stimulation?

We certainly agree that as the synaptic vesicles fuse, that there is less FFN to release with subsequent stimuli, and analysis of FFN destaining in previous papers has used this form of analysis. It is less relevant for FFN102 in the present approach, as we are taking advantage of the design feature that it becomes brighter upon release due to the shift in pH. This signal is measureable even following a short electrical stimulation, one second as compared with minutes for previous studies. The major consequence of release is therefore the transient which is a much larger signal, particularly in this very sparsely innervated region, in which the low signal of FFN within the acidic synaptic vesicles in thin axons is undetectable above background levels. Indeed, we now also show that in the GPe, there is no correlation between puncta and amount of release in Figure 7B.

To make the point more clearly that this novel approach takes advantage of a transient flash of release rather than the previous FFN approaches that analyzed destaining from axonal varicosities, we changed the title of the manuscript and state in the Abstract:

“we introduce an optical approach using a pH-sensitive fluorescent false neurotransmitter, FFN102, that exhibits increased fluorescence upon exocytosis from the acidic synaptic vesicle to the neutral extracellular milieu.”

In the Introduction:

“In this study, we adapt a pH-sensitive FFN, FFN102, which is a substrate for the dopamine transporter, DAT, and the vesicular monoamine transporter, VMAT2 (Rodriguez et al., 2013) as the first use of a “flashing FFN”. The signal from FFN102 is well suited for study synaptic release in sparsely innervated regions as it is brighter in the neutral extracellular milieu than the acidic milieu of the synaptic vesicle lumen (Lee et al., 2010).”

In the Results:

“We then addressed whether the differences in kinetics and size of FFN transients between the striatum and the GPe might result from differential reuptake, which might broaden the signal. […] Neither the decay time to its half-maximum value (Figure 6E) nor the shape of the decay were altered by nomifensine (Figure 6F), indicating a lack of a role for DAT in altering evoked FFN102 signals.”

And to discuss that the amount remaining in the axons is a less reliable indicator for very sparsely innervated regions we write:

“The greater release of FFN102 in GPe at higher frequencies (Figure 5) suggested that GPe dopamine axons might have the capacity to generate large FFN transients. […] Surprisingly, GPe regions with large FFN102 transients did not have obvious FFN102 puncta within the field of view (Figure 7B).”

In the Discussion:

“Here we introduce a new use for pH sensitive FFNs, which is to resolve release from dopamine axons in a region of very sparse innervation by “FFN flashes”, i.e., short duration calcium-dependent release transients during exocytosis as the fluorescence is unquenched upon exocytosis from the acidic vesicle lumen to the neutral extracellular space. FFN102, a DAT and VMAT2 substrate with a pK of 6.2 (Lee et al., 2010), is well suited for this application, as its signal increases nearly 400% between the acidic synaptic vesicle (~pH 5.6) and the extracellular milieu at pH 7.4 (Rodriguez et al., 2013).”

3) I wonder about the authors' claim that Ai9 ChR2 mice have been used, while no data is presented. The ChR2 mice have the potential to increase the signal-to-noise significantly.

We have deleted mention of the Ai9-ChR2 mice, as while the localization can be used, in our current approach, the LED light contaminated the FFN signal. We think that ongoing development of red-shifted photoactivatable channels by other research groups will be used for this purpose, but they are not yet available to us, and so we have no useful data on this approach.

4) What is the sensitivity of the permutation method to noise in the imaging results? I would think that if baseline recordings are noisy, then the permutation method may increase the fraction of tests that produce a "significant" answer. Is there a difference in the noise level in the various groups of experiments?

There is little difference in the noise level in the experiments, although there is far more baseline in the striatum than the GPe. We now analyze the data without the permutation method as written in the Materials and methods:

“To calculate baseline fluorescence, a linear fit was calculated to 500 ms of the data immediately prior to the stimulation. […] The smoothed baseline was used as the F for calculating ΔF/F for every time point of a given intensity.”

Other concerns this reviewer had with figures relating to hotspot analysis and modulation have been removed. A more detailed methodology has been used to assess hotspots. The possibility of measuring modulation is taken up in the Discussion.

Additionally, we have made a correction to the placement of the discussion related to Figure 6 (Critique 5).

5) The text refers to panels and data in Figure 6 (subsection “FFN102 release reveals differences in dopamine release between striatum and GPe”, first paragraph) that do not exist in the figure itself. This should be fixed before I can form an opinion on this data and its interpretation.

See response to comment #4.

Reviewer #2:

[…] 1) FFN102 loading results in a lot of signal outside of dopamine neurons (Figure 2E). This matches well with Figure 7, where little dopamine release is detected in areas with diffuse "loading", and this diffuse lower intensity signal is also present in the striatum (Figure 2E here and Figure 2 in PMID23277566). The most likely explanation is that there is background accumulation of the dye in non-dopamine neurons. This has to be discussed and better acknowledged in the text. Furthermore, Figure 2E overlap panels are very saturated in the blue=FFN channel, different from the non-merged images that show FFN only. More careful image editing, perhaps combined with a different color choice, should be applied.

This critique is related to Critique 2 by reviewer 1, please also see that discussion.

We have analyzed the relationship between “puncta”, i.e., static fluorescent areas and FFN release and find that in the very sparsely innervated GPe, there is no correlation between puncta and amount of release (Figure 7B).We now write: “The greater release of FFN102 in GPe at higher frequencies (Figure 5) suggested that GPe dopamine axons might have the capacity to generate large FFN transients. […] Surprisingly, GPe regions with large FFN102 transients did not have obvious FFN102 puncta within the field of view (Figure 7B).”

This may be related to the lack of significant synaptic vesicle clusters in this region of the axon, in very strong contrast to the same axons in the striatum, which have many distinctive presynaptic sites. We now write in the Discussion:

“On average, most GPe regions showed release after the first pulse, but the amount of evoked release was strikingly variable, and far more responsive to increased stimulus frequency than release in the striatum. […] The lower evoked FFN transients in GPe may be due to exocytosis from comparatively small reservoirs of synaptic vesicles in thin portions of the axon that do not maintain large vesicle clusters.”

We agree with the reviewer that diffuse loading is a concern, but note that 1) DAT blockade 2) the aphakia mutation 3) 6OHDA, nearly abolished FFN102 signal in both striatum and GPe. Thus, the non-dopaminergic component of signal in both brain regions is minor. We now write in the Results section:

“To examine if FFN102 is loaded in GPe axons as a substrate for the dopamine transporter (DAT), we co-incubated slices with both FFN102 and the DAT inhibitor, nomifensine, for 30 minutes before the stimuli. The significant decrease of evoked fluorescent transients in slices co-incubated with nomifensine is consistent with an inhibition of uptake and loading of FFN102 into DA axons (Figure 2C, D).”

“To confirm whether dopaminergic neurons are the source of the FFN transients, we used the toxin 6-hydroxydopamine (6-OHDA lesion) to unilaterally lesion the dopamine projections passing through the medial forebrain bundle. […] Thus, SNc dopamine neurons are the primary contributor to FFN102 release in the GPe.”

2) The experiments that address release characteristics and modulation are unclear at this point: a) The modulation experiments suggest that there is not the typical modulation of opioid and D2 receptors, and ACh receptor blockage perhaps mildly decreases the FFN102 signal during a stimulus train. However, the controls in the striatum suggest that FFN102 signals behave somewhat different from previous studies. The simple prediction from published studies would be that the first amplitude is smaller in MEC, but that dopamine release is rapidly depleted during 50 Hz/50 stimulus trains whether or not AChs are blocked. The method does not have time and spatial resolution to test this. Hence, it is possible that the differences observed have more to do with dye diffusion, clearance, etc., which could be different between the brain areas, than with release modulation.

b) Results describe experiments to suggest that "release differs between the GPe and the striatum" (section caption). Generally, better terminology to express the distinction between release and the FFN102 signal is needed. The signal is a function of release, diffusion, reuptake and reacidification. It is possible that differences in the signal arise because diffusion, reuptake etc. are different between the two brain areas, and release per se, from a bouton, is not different except that much less FFN102 is released because of the sparsity of dopamine terminals in GPe.

The meaning of these experiments is currently less clear than what the model shown in Figure 10 expresses. Leaving the data in the manuscript is fine to hint at potentially different modulation, but it may be better to remove the model and focus on the point that signals in an area with sparse dopamine innervation can be detected, rather than making strong mechanistic claims about the release.

We agree with these critiques, and have deleted the model and analysis of modulation. We further agree that as the reviewer writes, a better terminology is needed. We now use the term “flash”, i.e., the evoked FF102 transient. As discussed in the DAT blockade experiments, the flash is now more thoroughly explained as being due to release and the consequent increase in emissia due to the pH shift, and the decay is due to diffusion, whereas reacidification and reuptake do not play significant roles during the rapid transient events.

3) The calcium dependence is shown for N = 1 slice in Figure 3C, D, this is simply below acceptable standards.

We have repeated all of the calcium experiments with N=9 slices in the striatum and N=16 slices in the GPe. We performed within-slice comparisons at three concentrations to obtain data for these experiments, which are in the revised Figure 3.

4) In Figure 7, it would be more meaningful to show a correlation between the morphological appearance (panel A) and the FFN transient (panel B). As shown, Figure 7 does not establish that areas with sparse but highly fluorescent axons release more. Furthermore, the "high" panels in A poorly reflect that point compared to the "low" panels. A better analysis is necessary.

We have followed the reviewer’s advice and added significant new analysis, and indeed the more sparsely innervated regions with highly fluorescent axons do not release more. As discussed above, the low level of signal in the synaptic vesicle prior to release means that the flash is more important for this analysis than the background levels. This is we think a facet of measuring in very sparsely innervated regions. The new data are in Figure 7. We now write:

“The greater release of FFN102 in GPe at higher frequencies (Figure 5) suggested that GPe dopamine axons might have the capacity to generate large FFN transients. […] We then determined each field of view’s derivative value before stimuli, after a single pulse, and after two pulses (Figure 7H). For both the striatum and the GPe, the largest transient occurred at the first electrical pulse.”

[Editors' note: the author responses to the re-review follow.]

Major comments:

1) All the illustrations used multiple stimuli to evoke release and the rise in fluorescence is slow and sustained. Single stimuli show rapid rise and fall of dopamine using both voltammetry and the new sensors. Thus the present results appear to be quite different. The authors should isolate responses to single stimuli (for instance by initially using sparse stimulation), followed by strong stimulation to identify the events and loci.

We believe that the impression of apparently slow kinetics of the dopamine signal and differences between single and train stimuli is a misunderstanding and try to correct it here: the FFN102 signal rise/release is not slower than results using cyclic voltammetry in the striatum, and in the GPe is not sustained.

As shown in Figure 1 of (Schmitz et al., 2001)], with voltammetry, the rise in the striatum occurs at about 180 msec (dotted line in the figure), and the sampling using this technique is typically 10 Hz, which is at or near the effective limit due to the charging “Faradaic” current which produces excessive background at higher charging rates. In this older study from our group, we showed that the cyclic voltammetry mode of electrochemistry is more accurate for determining release and reuptake than amperometry, as the second approach consumes dopamine at the electrode surface that creates a local concentration minimum and removes local extracellular dopamine faster than the DAT, whereas cyclic voltammetry regenerates the dopamine, and neither consumes it nor interferes with the kinetic analysis of DAT. The consuming face also artificially decreases the rise time of the signal by vastly increasing the local concentration gradient. Therefore, the estimate of ~180 msec for the rising phase of dopamine by a single pulse in the striatum is to our knowledge the best estimate of response to evoked release by electrical stimuli with extant techniques. Note also that the 180 msec is far longer than the release from synaptic vesicle fusion which is less than 1 msec (Staal et al., 2004), and is due to the overflow/diffusion of dopamine from multiple release sites to the carbon fiber: this is however the situation for dopamine neurotransmission in the striatum and other regions as the dopamine receptors are extrasynaptic and are activated at this time scale.

This means that with FFN102, the 10 pulse 10 Hz stimuli and measurements at 10 Hz that the rise time is essentially identical to cyclic voltammetry. This may be clearest to see in the present study’s Figure 3, which shows a rapid increase in slope by the second pulse for both the striatum (A) and GPe (C), i.e., by 200 msec, and that with the sampling we are using, it is at the limit of resolution. It is true that FFN optical signals do not have the excellent level of low noise and signal resolution of carbon fiber electrochemistry, but the GPe signal is too small for electrochemistry to measure at all. The FFN signal is in contrast sufficient to clearly indicate that the rise time is also about 200 msec, essentially identical to voltammetry.

We also analyze the kinetics of the rising phase by reporting the derivative, i.e., the slope, in Figure 7H that shows that both for striatum and GPe, nearly all of the rising phase is due to the release from the first stimuli. Figure 7G shows a higher derivative of the first pulse from the striatum than GPe, but this is because as the signal is higher in the striatum, the rising slope is as well. It may be that the impression that this is a slow rise is because we did not make it sufficiently clear that with the spiral scan used in this study, each data point requires 100 msec, that is we can only measure a change at 10 Hz intervals. This is quite rapid for optical methods in the brain slice, and very slow for electrophysiology, and the same sampling rate used for cyclic voltammetry.

We trigger the start of the image acquisition and the first pulse simultaneously, and so the response to the first pulse is complete only during the second point, when the spiral scan returns to the start point. The rise time may be faster than 100 msec following the stimulus, but the limited effective sampling of the spiral scan in the slice preparation, in comparison to electrophysiology, limits us to resolve to that duration.

Finally, the rationale for using trains of pulses rather than a single pulse, even though most of the release occurs in response to the first pulse is because of the high noise in optical measurements (in comparison to electrochemistry). With the pulse train, the signal remains elevated for over a second, which provides 10 data points, and this helps to determine that a “random” fluctuation in the light level is not misinterpreted as an FFN signal.

In the Materials and methods, we had previously written and retain the statement:

“Images.. were recorded at 10 Hz (6 µs dwell time) using a spiral scan.”

and now add:

“The use of a stimulus train provides a longer duration optical signal, and so a more robust detection of FFN release events by distinguishing the signal from rapid fluctuations.”

To make it clear that there is a rapid increase in the rise time, we now write:

“While we only resolve signals at 100 msec intervals, it is apparent that for both the striatum and the GPe, the largest contribution to the total transient occurred at the first electrical pulse and is consistent with estimates of the rising phase using cyclic voltammetry, which is ~180 msec in the striatum (Schmitz et al. 2001).”

Regarding the kinetics of the signal decay, please note that the decay in Figure 1A is much shorter for the GPe than for the striatum, and that in Figure 2C, the falling phase is complete within a second after the stimulus: if we could resolve it better, we suspect that it would be well fit by an exponential decay, consistent with diffusion from a point source. As we hope to have made clear in the present paper, dopamine in the striatum is mostly cleared by DAT, and while FFN102 is a DAT substrate it is of much lower affinity than dopamine itself and not rapidly cleared, so that diffusion plays a bigger role. We believe that the reliance on diffusion for the decay of the FFN102 signal is confirmed in Figure 6, which shows that nomifensine does not increase the duration of the signal. In the GPe, there is very little DAT activity as the dopamine axons are very sparse, and so diffusion plays a greater role: however, there is also less FFN released, and so the signal is of much shorter duration than for the striatum. We note:

“The longer fluorescence decay in striatum is in part due to diffusion from out-of-plane striatal terminals, whereas in the GPe, areas of high release are few and spatially separated, and so less signal would diffuse from distal release sites to the regions of interest (Sulzer and Pothos, 2000).”

In sum, we hope that by now clearly stating that the kinetics of the rising phase of the FFN signal is consistent with that of cyclic voltammetry that the issue is more completely analyzed.

2) One conclusion was that the fluorescent increase in the globus pallidus was smaller and declined more rapidly. This is not surprising given that the experiments were done at room temperature. If the reuptake process plays any role in the decline in fluorescence would be inactive so the decline would be strictly dependent on diffusion. It could be that the lower concentration of FFN102 measured in the GP dropped below the detection limit thus appearing to be more rapid. Even in the hot spots of the GP the diffusion and dilution away from the point is most likely faster than might be expected from what is observed from multiple release sites in the striatum.

This critique is very closely related to above discussion. We agree with all of these points: diffusion indeed appears to be responsible for the rate of the falling phase, and this is experimentally consistent with the nomifensine experiments in Figure 6, demonstrating that under these conditions that there is very little effect of DAT reuptake on FFN102. We further agree that there is indeed more FFN102 released from the multiple sites in the striatum, and that this is why diffusion appears “faster” in the GPe than striatum: more accurately, diffusion is a rate δC (concentration)/δx (distance) governed by the concentration gradient, not a velocity and so “faster” is a misnomer. We of course agree that the signal decay is indeed faster and that the higher concentration in the striatum leads to an apparently much “slower diffusion” to reach the detection limit.

We also agree with the implication that it is preferable in principle to perform experiments at physiological temperature, but the motion artifacts preclude this with our current techniques: we hope that we and the field will be able to overcome this limitation. To acknowledge the point about the contribution of diffusion to the decay phase of the signal in the GPe vs. striatum, in addition to the statement to Critique 1, we write:

“The decay of FFN102 transients below the apparent detection limit was much faster in the GPe than the striatum, including in GPe hotspots with high evoked transient amplitudes similar to the striatum. […] The longer fluorescence decay in striatum is in part due to diffusion from out-of-plane striatal terminals, whereas in the GPe, areas of high release are few and spatially separated, and so less signal would diffuse from distal release sites to the regions of interest (Sulzer and Pothos, 2000).”

3) The difference in the results obtained with the paired stimulation in the GP and dorsal striatum could be accounted for by the presence of the cholinergic interneurons in the striatum that increase the probability of dopamine release dramatically such that there is substantial paired pulse depression of dopamine release.

We agree in part, and the reviewer is likely referring to multiple studies by Stephanie Cragg’s group, Joseph Cheer, as well as a study by Melchoir et al. from Sara Jones’s lab (2015) demonstrating that within the striatum, activation of cholinergic receptors dramatically enhances dopamine releases after a single pulse of electrical stimulation (Rice and Cragg, 2004; Cachope et al., 2012; Melchior et al., 2015). Please note that the Melchoir et al. study showed that a train of 20 electrical pulses, similar to the 10 pulse protocol we employ here, produced the same amount of dopamine release whether or not DHbE was present. This also relates to an early study by our group showing that the effect of nicotine and antagonists on evoked dopamine release are overcome by higher frequency activity (Zhang and Sulzer, 2004), and independently by Rice and Cragg (Rice and Cragg, 2004). To our knowledge, all of the extant work in the field indicates that cholinergic receptors exerts relatively little effect on the total amount of dopamine released during a train of pulses.

We now write:

“Local interactions with cholinergic interneurons may contribute to differences in the FFN signals between striatum and GPe, although these effects are minimized with the train stimuli used in this study (Melchoir et al. 2015, Zhang and Sulzer 2004, Rice and Cragg, 2004).”

4) There was no evidence for recovery (decrease) in fluorescence of FFN102 after cessation of stimulation in the striatum (see Figure 1). Therefore, the discussion of the authors concerning the kinetics of recovery doesn't appear to be consistent with the data (Discussion, fourth paragraph).

Thank you for pointing this out, we had given a mistaken impression. There is a recovery of fluorescence of FFN102 in the striatum, but it is very slow, as shown here from Figure 5 of Rodriguez et al., 2012. We now revise the former statement to read:

“The decay of FFN102 transients below the apparent detection limit was much faster in the GPe than the striatum, which requires several minutes (Rodriguez et al., 2012) including for the GPe hotspots with high evoked transient amplitudes similar to the striatum.”

5) The idea of "hot spots" of release: The data presented in Figure 7 are not convincing. The authors explain that data were measured as an average of fields of view. How does this relate to the idea of hotspots? Especially when the authors refer to structures which should be of smaller dimensions than the fields. For example, in the Introduction, the authors write: "The identification of hotspots in the GPe also supports recent work showing that many striatal dopamine varicosities with clusters of synaptic vesicles are silent (Pereira et al., 2016), and this could be due to the absence or presence of local presynaptic scaffolding proteins (Liu et al., 2018)." However, the dimensions of the so-called "hot spots", as measured in the current study, are significantly larger than what one would expect varicosities to be. Therefore, such a suggestion as to the biological basis for the hotspots does not appear plausible. Later, the authors write: "For example FFN transients could be used to locate dopamine release near specific cell types, such as the arkypallidal cells which project back into the striatum and the protopallidal cells which project downstream to the SNr and GPi (Gittis et al., 2014; Mastro et al., 2014; Hernández et al., 2015)." Are the dimensions of the hotspots small enough to find specific cells within slices? If the authors measure transients from whole fields, then this does not appear to be the case.

Thank you for this comment, and we were not sufficiently clear in the previous version. We indeed analyze the increased fluorescent signal over an entire field of view, which is 30 x 30 µm: presynaptic varicosities in the striatum labeled with FFN are close to 1-2 µm, and the field of view is close to the scale of one or two cell bodies. Therefore, we can only provide an upper limit of a “hot spot” in a region.

Nevertheless, please note that in Figure 7E, fields of view with high release (yellow circles) are neighbored by and even overlap with fields of view with dramatically less release in a representative slice. This is analyzed in Figure 7F, which shows that the spatial distribution of release sites was not concentrated in clusters of “hotspots”, and that they show no more tendency to be close together than other sites. Thus, a “hotspot” in one field of view did not make it any more likely to find one in a nearby field of view. Given that we do localize “hotspots” to regions the size of 1-2 cell bodies, we think this is small enough to be close to particular cells within the slice or near specific populations of dendrites. To make this point more clearly, we now write:

“The current approach is able to resolve FFN transients in 30 x 30 µm fields of view, close to the size of neuronal cell bodies, and so FFN transients might be used to characterize dopamine release near specific cell types, such as the arkypallidal cells which project back into the striatum and the protopallidal cells that project to the SNr and GPi (Gittis et al., 2014; Mastro et al., 2014; Hernández et al., 2015).”

https://doi.org/10.7554/eLife.42383.020

Article and author information

Author details

  1. Jozsef Meszaros

    1. Laboratory for Functional Optical Imaging, Department of Biomedical Engineering, Mortimer B. Zuckerman Mind Brain Behavior Institute, Columbia University, New York, United States
    2. Graduate Program in Neurobiology and Behavior, Columbia University, New York, United States
    Contribution
    Conceptualization, Data curation, Formal analysis, Validation, Investigation, Visualization, Methodology, Writing—original draft, Writing—review and editing
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0002-2485-0144
  2. Timothy Cheung

    Department of Neurology, Columbia University, New York, United States
    Contribution
    Conceptualization, Investigation, Writing—review and editing
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0002-7516-8321
  3. Maya M Erler

    Graduate Program in Pharmacology, College of Physicians and Surgeons, Columbia University, New York, United States
    Contribution
    Conceptualization, Data curation, Formal analysis, Investigation, Visualization, Writing—review and editing
    Competing interests
    No competing interests declared
  4. Un Jung Kang

    Department of Neurology, Columbia University, New York, United States
    Contribution
    Resources
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0002-5970-6839
  5. Dalibor Sames

    Department of Chemistry and NeuroTechnology Center, Columbia University, New York, United States
    Contribution
    Resources
    Competing interests
    No competing interests declared
  6. Christoph Kellendonk

    1. Division of Molecular Therapeutics, New York State Psychiatric Institute, New York, United States
    2. Department of Psychiatry, Columbia University, New York, United States
    3. Department of Pharmacology, Columbia University, New York, United States
    Contribution
    Conceptualization, Resources, Supervision
    For correspondence
    ck491@cumc.columbia.edu
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0003-3302-2188
  7. David Sulzer

    1. Department of Neurology, Columbia University, New York, United States
    2. Division of Molecular Therapeutics, New York State Psychiatric Institute, New York, United States
    3. Department of Psychiatry, Columbia University, New York, United States
    4. Department of Pharmacology, Columbia University, New York, United States
    Contribution
    Conceptualization, Resources, Supervision, Funding acquisition, Methodology, Writing—original draft, Writing—review and editing
    For correspondence
    ds43@cumc.columbia.edu
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0001-7632-0439

Funding

National Institutes of Health (T32 NS06492B-04)

  • Jozsef Meszaros

Parkinson's Disease Foundation

  • Timothy Cheung
  • Un Jung Kang
  • David Sulzer

National Institute of Neurological Disorders and Stroke (R01 NS101982)

  • Un Jung Kang
  • David Sulzer

U.S. Department of Defense (PR161817)

  • Un Jung Kang

National Institute of Neurological Disorders and Stroke (R03 NS096494)

  • Un Jung Kang

National Institute of Mental Health (R01 MH108186)

  • Dalibor Sames
  • David Sulzer

National Institute of Mental Health (RO1 MH093672)

  • Christoph Kellendonk

JPB Foundation

  • David Sulzer

National Institute on Drug Abuse (R01 DA07418)

  • David Sulzer

The funders had no role in study design, data collection and interpretation, or the decision to submit the work for publication.

Acknowledgements

We thank M Sonders for expert technical assistance. We also thank Y Schmitz for crucial advice about dopamine release measurements in the striatum, M Dunn for important discussions about FFNs, Y Schmitz and M Dunn for comments on an earlier version of this paper, and Ethan Bromberg-Martin for useful discussions about data analysis.

Ethics

Animal experimentation: All animal protocols followed NIH guidelines and were approved by Columbia University's Institutional Animal Care and Use Committee.

Senior Editor

  1. Eve Marder, Brandeis University, United States

Reviewing Editor

  1. Inna Slutsky, Tel Aviv University, Israel

Version history

  1. Received: October 9, 2018
  2. Accepted: December 18, 2018
  3. Accepted Manuscript published: December 19, 2018 (version 1)
  4. Version of Record published: January 8, 2019 (version 2)

Copyright

© 2018, Meszaros et al.

This article is distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use and redistribution provided that the original author and source are credited.

Metrics

  • 1,726
    Page views
  • 271
    Downloads
  • 10
    Citations

Article citation count generated by polling the highest count across the following sources: Crossref, PubMed Central, Scopus.

Download links

A two-part list of links to download the article, or parts of the article, in various formats.

Downloads (link to download the article as PDF)

Open citations (links to open the citations from this article in various online reference manager services)

Cite this article (links to download the citations from this article in formats compatible with various reference manager tools)

  1. Jozsef Meszaros
  2. Timothy Cheung
  3. Maya M Erler
  4. Un Jung Kang
  5. Dalibor Sames
  6. Christoph Kellendonk
  7. David Sulzer
(2018)
Evoked transients of pH-sensitive fluorescent false neurotransmitter reveal dopamine hot spots in the globus pallidus
eLife 7:e42383.
https://doi.org/10.7554/eLife.42383

Further reading

    1. Neuroscience
    Amanda J González Segarra, Gina Pontes ... Kristin Scott
    Research Article

    Consumption of food and water is tightly regulated by the nervous system to maintain internal nutrient homeostasis. Although generally considered independently, interactions between hunger and thirst drives are important to coordinate competing needs. In Drosophila, four neurons called the interoceptive subesophageal zone neurons (ISNs) respond to intrinsic hunger and thirst signals to oppositely regulate sucrose and water ingestion. Here, we investigate the neural circuit downstream of the ISNs to examine how ingestion is regulated based on internal needs. Utilizing the recently available fly brain connectome, we find that the ISNs synapse with a novel cell-type bilateral T-shaped neuron (BiT) that projects to neuroendocrine centers. In vivo neural manipulations revealed that BiT oppositely regulates sugar and water ingestion. Neuroendocrine cells downstream of ISNs include several peptide-releasing and peptide-sensing neurons, including insulin producing cells (IPCs), crustacean cardioactive peptide (CCAP) neurons, and CCHamide-2 receptor isoform RA (CCHa2R-RA) neurons. These neurons contribute differentially to ingestion of sugar and water, with IPCs and CCAP neurons oppositely regulating sugar and water ingestion, and CCHa2R-RA neurons modulating only water ingestion. Thus, the decision to consume sugar or water occurs via regulation of a broad peptidergic network that integrates internal signals of nutritional state to generate nutrient-specific ingestion.

    1. Neuroscience
    Lucas Y Tian, Timothy L Warren ... Michael S Brainard
    Research Article

    Complex behaviors depend on the coordinated activity of neural ensembles in interconnected brain areas. The behavioral function of such coordination, often measured as co-fluctuations in neural activity across areas, is poorly understood. One hypothesis is that rapidly varying co-fluctuations may be a signature of moment-by-moment task-relevant influences of one area on another. We tested this possibility for error-corrective adaptation of birdsong, a form of motor learning which has been hypothesized to depend on the top-down influence of a higher-order area, LMAN (lateral magnocellular nucleus of the anterior nidopallium), in shaping moment-by-moment output from a primary motor area, RA (robust nucleus of the arcopallium). In paired recordings of LMAN and RA in singing birds, we discovered a neural signature of a top-down influence of LMAN on RA, quantified as an LMAN-leading co-fluctuation in activity between these areas. During learning, this co-fluctuation strengthened in a premotor temporal window linked to the specific movement, sequential context, and acoustic modification associated with learning. Moreover, transient perturbation of LMAN activity specifically within this premotor window caused rapid occlusion of pitch modifications, consistent with LMAN conveying a temporally localized motor-biasing signal. Combined, our results reveal a dynamic top-down influence of LMAN on RA that varies on the rapid timescale of individual movements and is flexibly linked to contexts associated with learning. This finding indicates that inter-area co-fluctuations can be a signature of dynamic top-down influences that support complex behavior and its adaptation.